Understanding Accruals, Researching investors in Dutch listed companies : the blindness of investors

(1)

Amsterdam Business School

Understanding Accruals, Researching investors in Dutch listed

companies

The blindness of investors

Name: Tim Smal

Student number: 10681183

Supervisor: dr. S.W. (Sanjay) Bissessur

Second supervisor : dr. lr. S.P. (Sander) van Triest Date: January 24, 2016

Paper: Master thesis Version: Final version Word count: 14.418

MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam

(2)

Statement of Originality

This document is written by Student Tim Smal who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

Abstract

This paper investigates whether the quality of the financial statements in Dutch listed companies is high by researching which investment choice by which group of investors is causing the accrual anomaly to exist in the Dutch market. Results show that the quality of the financial statements is not high. This is because investors, the most important users of the financial statements according to the IASB, are not reacting in the right way to the accrual information available. However, unlike earlier research in other economic markets, my findings show that accrual anomaly is caused by institutional investors investing in low accrual firms.

(4)

1 Introduction

Investors tend to react to every signal that listed firms provide them with, because a signal can lead to an increase or decrease in stock prices. By responding to the given information in a timely matter, investors are gaining profits. However, research has shown that investors are still not fully reacting to all the information presented. One of the phenomenon’s in economics today shows that investors do not understand the accruals in the financial statements. This causes them to undervalue or overvalue companies by misreading the future earnings information presented to them via accruals. This phenomenon is known as the accrual anomaly (Battalio et al., 2012; Bradshaw, 2001; Collins et al., 2003; Galanou, 2012; Hirshleifer et al., 2012; Pincus et al., 2007; Sloan, 1996).

Sloan (1996) was the first one to report about the accruals anomaly. He showed that investors fixate on the firms’ earnings rather than distinguishing the accrual component and the cash flow component of the current earnings. This was an unexpected finding because both the accrual component as the cash flow component are important components that can provide the investor information about future earnings. The paper of Sloan (1996) was a major contribution to financial research, which today reports an abundance of research on the accrual anomaly.

The accrual anomaly is the term for stock prices that seem to fail to incorporate information about the accruals levels of firms (Battalio et al., 2012) which leads to over- and undervaluation of firms, and is one of the contradictions of the efficient market hypothesis (Sawicki & Shrestha, 2011). Lev & Nissim (2006) describe the accrual anomaly as: “The negative relationship between accounting accruals and subsequent stock returns” (p.193). By investing short in companies with high accruals and long in companies in low accruals companies Sloan (1996) found that it is possible to get up to 10,4% profit on a stock portfolio by correct interpretation of the accruals in financial statements.

The efficient market hypothesis states that all publicly available information is incorporated in the stock prices, making the financial markets efficient. To decrease the accrual anomaly, investors need to react to accrual information at the time this information is presented. But this seems rather difficult for investors, according to what research on the accrual anomaly has showed us the last years. One of the reasons that accruals are not well understood by investors with regards to the prediction of future earnings of firms they might invest in, is that the accrual component of the financial statement is easier to manipulate relative to the cash flow component (Sloan, 1996; Galanou, 2012; Leippold & Lohre, 2012). A critical question remains however, how it is possible that given all of the research on the accrual anomaly, this phenomenon remains?

(6)

In this research I will be looking at specific groups of traders, trying to find evidence which investors seem to react to accruals. This way I want to provide evidence about which group of investors cause accrual anomaly. Furthermore I want to extend this research by comparing the two groups of investors. By doing this I will be able to tell which investors cause the accrual anomaly and if it is due to investment decisions in high accrual companies or due to decisions in low accrual companies.

Battalio et al. (2012) investigated different groups of investors, and results show that investors that initiate small trades tend to respond in the opposite direction of the accrual information, suggesting that these investors are attracted to “attention grabbing” stocks, while also showing that large traders seem to act correctly to the accrual information. In my research I want to further look at groups of investors, trying to provide evidence for other groups of investors and thus trying to provide evidence about the group of investors that cause the accrual anomaly. Also I will be focusing on the Dutch listed companies, because I want to investigate whether findings from research in the U.S. are also applicable to, in this case, the Dutch market.

While Galanou (2012) has already provided evidence about the existence of the accrual anomaly in the Dutch stock market, looking at stock listed Dutch companies during the time period 1988 – 2007, I want to focus on data during the time period 2005 – 2014, and

furthermore, I will be looking at different groups of investors to provide more evidence about the investors causing accrual anomaly.

By focusing on the Dutch market I will be providing external validity for the research of Battalio et al. (2012) as well as for the research of Sloan (1996) by answering the research question: ‘Do small investors in Dutch listed companies cause accrual anomaly?’. Due to earlier research done on groups of investors that state that the bigger investors (mostly institutional investors) tend to react in the right way to the accrual information (Battalio et al., 2012; Collins et al., 2003), this research question is stated in the direction of the smaller investors. While Battalio et al. (2012) researched whether small or big investors are the cause of the accrual anomaly, this research will not only focus on the existence of the accrual anomaly and by which group it is caused, but this research furthermore investigates whether the accrual anomaly is existing due wrong investment choices in firms with low accruals or due to wrong investment choices in firms with high accruals. By focusing on the financial statements I will be able to conclude whether the quality of the financial statements is up for change.

Contribution to accounting can be found in research of the Dutch economic market, by measuring the quality of the financial statements using the quality of information that the

(7)

accruals provide, and by searching for differences in investment decisions between the portfolios of institutional investors and small investors.

(8)

2 Literature

In the majority of the research conducted by accounting students, as in this paper, the research is focused on the reporting quality and more specific, on how inconsistencies in accounting can be examined. Furthermore, this research can be expanded to how reporting quality can be improved by examining the cause of these inconsistencies. I will focus on how users of the financial

statements react to specific information in the financial statements. Reporting quality will be measured using accruals, a measurement commonly used in research to indicate reporting quality (Sloan, 1996; Schipper & Vincent, 2003; Dechow et al., 2010). While looking at the users of the financial statements, I will be focusing on investors and on the way they interpret accruals in financial statements. By looking at reporting quality through the eyes of users of the financial statements, I will try to provide additional evidence on the accrual anomaly by combining data about the volume of stock traded with the accrual information and by comparing the investment choices of institutional investors with the investment choices of small investors.

In the first paragraph of this chapter reporting quality and accounting quality are

explained. The second paragraph explains the different kind of investors. In this paragraph the group of financial statement users that this research fixates on will be described. The third paragraph explains the term ‘Earnings quality’. This term is explained due to the importance of earnings quality for investors in financial statements. The fourth paragraph explains accruals and describes how accruals in financial statements occur. The last paragraph of this chapter is focused on the accrual anomaly, the phenomenon that occurs when investors do not fully understand the accruals in financial statements, resulting in incorrect estimation of future profits. By using this structure for the literature chapter, I try to start of high in theory while descending and delimiting on just a small part of the accounting literature, the accrual anomaly.

2.1 Reporting Quality & Accounting Quality

Reporting quality, as described by the IASB (International Accounting Standards Board), has two primary qualitative characteristics of information in the financial statements. These are:

Relevance and a Faithful representation (Palea, 2013). Without relevant and faithful information, the financial statements are considered to be useless to the users of the financial statements. Information in financial statements is considered relevant when the information presented is able to make a difference to the user of the financial statements. Faithful representation means that information in the financial statements reflect the real-world economic phenomena that the statements purports to present (Palea, 2013). Besides these two qualitative characteristics there are enhancing qualitative characteristics that complete the fundamental characteristics. These are:

(9)

comparability, verifiability, timeliness and understandability. When information is lacking on one of these four qualitative characteristics, information is considered to be less useful information.

Both the IASB Framework and the FASB Concepts Statements identify comparability as “the quality of information that enables users to identify similarities in differences between two sets of economic phenomena”. This means that the financial statements have to be consistent throughout the years and financial statements of different companies have to be drafted in the same format. Due to this qualitative characteristic, investors can compare different companies and make the best investment choice. Verifiability, as described by the conceptual framework, means that various different and independent observers could agree that a specific representation is a faithful representation. Timeliness describes the factor of timing within financial reporting. When information in the financial statements is recognized timely, users of the financial statements can use this information. If information is recognized half a year after year-end, this information is mostly less useful due to the timing of the recognition. Timeliness however also can be interpreted as the publishing date of the financial statements. If the financial statements are published eleven months after the ending of the fiscal year, this information is less useful to users of the financial statements. Understandability is described in the conceptual framework of the IASB to be the term for clear and concise information (Nobes & Stadler, 2014). If

information needed by the users of the financial statements is ‘hidden’ somewhere within in the financial statements, this makes the information less useful.

Useful information in the financial statement forms the basis for the conceptual framework of the IASB (2010 CB 1.16). This means that the conceptual framework has its primary focus on providing useful information to creditors, investors and other people that form and make decisions based on the financial statements. The most important group of financial statements users are ‘the investors’. This group will be further described in the next paragraph. Guidelines of the IASB are focused on the investors because investors, according to the IASB, need the information from financial reports the most. An important argument the IASB gives is that investors cannot get important information about the future benefits of the firm directly from the firms.

Barth et al. (2008) examined whether the adoption of the International Accounting Standards (IAS) leads to higher accounting quality. The goal of the IASB is to improve the reporting quality by IAS. Barth et al. (2008) findings are consistent with the goal the IASB described. By examining firms in 21 countries Barth et al. (2008) finds evidence that by applying IAS, firms apply more timely loss recognition, less earnings management and more value

(10)

quality is the term for the quality of the financial statements as presented to users of the financial statements, accounting quality describes the quality of the financial statements due to accounting policies within firms. Therefore, in order to conclude something about reporting quality, it is important to research the accounting quality. Because the reporting quality is affected by the underlying accounting quality.

Accounting quality, is measured by Barth et al. (2008) by using earnings management, value relevance and timely loss recognition metrics. These are formulas used in earlier research that will test whether earnings management is lower, whether there is a more timely loss recognition and whether the value relevance is higher after implementation of IAS.

Earnings management is also known as the manipulation of the firms’ earnings, as reported in the financial statements (Hadani et al., 2011), and is mostly measured by analyzing the accruals (Hadani et al., 2011). One of the most commonly used models is the modified Jones model (1991) (Dechow et al., 2010; Hadani et al., 2011; McNichols, 2002). And one of the most common causes for earnings management is the bonus of the manager (Hadani et al,. 2011; Healy, 1985; Jones, 1991), this is because bonus schemes create incentives for managers to increase or decrease earnings.

Information is considered to be value relevant when it has a predicted association with equity market values (Barth et al., 2001). Barth et al. (2008) finds that the implementation of IAS provides the user of the financial statements with higher value relevance of net income for good news stock returns. Value relevance can be measured by testing the explanatory power from a regression of stock price or by testing the explanatory power from regressions of net income per share on annual stock return (Barth et al., 2008).

A timely loss recognition means that gains and losses are accounted for timely. Timely in this case means the loss recognition must occur around the time of revisions in expectations of future cash flows (A& Shivakumar, 2006). These revisions are done by adjusting accruals (Ball & Shivakumar, 2005; Ball & Shivakumar, 2006).

Thus, by focusing on accruals in financial statements and by focusing on the stock returns of the companies, I will be able to draw a conclusion about both the accounting quality and the reporting quality of the Dutch economic market. Beatty et al. (2010) state that when financial reporting quality is higher, the information asymmetry is lower, and when accounting quality is high, the reduction in information asymmetry leads to an improvement in investment. This implies that investors will invest more when the accounting quality and the reporting quality is higher. Findings of Bushman et al. (2011) and Biddle et al. (2006) are consistent with the findings of Beatty et al. (2010). Bushman et al. (2011) finds that higher accounting quality leads to

(11)

increased investment efficiency, and Biddle et al. (2006) show that higher accounting quality is associated with lower investment cash flow sensitivity.

In the next paragraph the most important group of users of the financial statements according to the IASB, the investors, will be described.

2.2 Investors

People or companies that buy stock from other companies, making them shareholders in the company, are considered to be investors. However, stock is not the only thing people want to invest in. Investors can also be individuals or big companies that are buying and selling

properties such as houses and commercial properties. Due to the fact that this research focusses on the stock market, the investors that buy and sell stock on the stock market will be described.

There are different types of investors. The biggest group of investors are called institutional investors. Institutional investors are large organizations such as insurance companies, pension funds, banks and finance companies which invest large sums of money. Institutional investors are, at this moment in time, a major component of equity markets in many Anglo-American countries (Wang, 2014). Due to the size and amount of money of this group of investors they have an information advantage on the smaller investors. But only the largest institutional investors within the group of institutional investors have these informational advantages (Schnatterly et al., 2008). Information advantages are present due to relationships with firms and due to active following of firms (Schnatterly et al., 2008). Furthermore, because institutional investors monitor the companies in which they invest better than the smaller investors, earnings management at the companies they invest in descends (Hadani et al., 2011).

Another group of investors is the small investors. These investors are mostly individuals looking for profits in the stock market by buying and selling small amounts of shares. These small investors, or individual investors as Hoffman & Seffrin (2014) describe them, are the investors that are ignored in science (Hoffman & Seffrin, 2014), because most of the research is based on the institutional investors. Small investors tend to invest in lottery-like securities (Kumar, 2009; Hoffman & Seffrin, 2014). These securities feature high risk and feature high negative risk-adjusted returns. Major difference between small investors and institutional investors is that institutional investors adjust the analysts’ recommendations downwards while small investors take these recommendations literally (Malmendier & Shanthikumar, 2007). This conclusion in the research of Malmendier & Shanthikumar (2007) states that institutional investors act rather conservative in regards to the smaller investors. Final conclusion of their research is that small investors trust analysts too much and are therefore considered to be naive

(12)

(Malmendier & Shanthikumar, 2007). Shanthikumar (2012) finds that smaller investors react differently to earnings surprises than the institutional investors do. When earnings surprises are negative, the smaller investors react more negatively to this information than the institutional investors, and when earnings surprises are positive, the smaller investors react more positive than institutional investor.

As described before, investors are only willing to buy shares if they see a future profit in that specific investment. Therefore it is important to know which information the investors use to make their decision to buy or sell shares.

Investors rely on the information provided by management of the firms (Hadani et al., 2011) and investors tend to look at a lot of information the company provides, before they are willing to buy shares. This for instance can be information about the signing of a new CEO that has already showed that he or she is able to increase the profitability of a company, or

information about a possible takeover. If investors see this new CEO as the person that can realize higher future earnings, they will tend to invest more. And if the takeover will most probably generate higher or more future earnings, investors will be eager to invest as well. But firms can also provide investors with information about future earnings by publishing earnings announcements or information from stock analysts. Due to the nature of this research, which focusses on the reporting quality of the financial statements, it is important to know which information investors look for in the financial statement. As described earlier, investors are looking at information about future earnings, therefore it is important to know what it is in the financial statements, that provides the investor with the information they need.

Future earnings information in the financial statements can be gained from the accrual and cash flow components of current earnings (Bradshaw et al., 2001; Sloan, 1996). Earnings can be divided in an accrual component and a cash flow component. The cash flow component shows the earnings from cash, generated in a specific year, where the accrual component shows the earnings of that same year, only this earnings are not cashed yet. The debtors are the best example of an accrual that has led to revenue but has not led to cash yet. Paragraph 2.4 explains the accruals.

Studies about future earnings information (Cotter, 1996; Boubakri, 2012) highlighted the importance of accruals and cash flow components of current earnings. Due to the importance of the quality of earnings information, the next paragraph focusses on the earnings quality within the financial statements. In the fourth paragraph, and exhibit 1, the cash flow and accrual components are explained by using an example used by Dechow et al. (2011).

(13)

2.3 Earnings Quality

Earnings quality is fundamental for the performance of a firm (Dechow et al., 2010). As described in paragraph 2.1 earnings quality is part of the overarching reporting quality and accounting quality. Dechow et al. (2010) defines earnings quality as follows: “Higher quality earnings provide more information about the features of a firm’s financial performance that are relevant to a specific decision made by a specific decision-maker.” (p.2). This definition in Dechow et al. (2010) is taken from SFAC No.1. By reading the definition it becomes clear that that specific decision-maker, as earlier mentioned in the first paragraph, is the investor.

Schipper & Vincent (2003) consider earnings quality to be constructs derived from the time-series properties of earnings, qualitative characteristics from the conceptual framework, the relations between cash, accruals and income, and implementation decision. Dechow et al. (2010) agree with this term of earnings quality, but they specifically zoom in on the qualitative

characteristics of reporting quality. According to Dechow et al. (2010), earnings quality is about decision usefulness of the information shown, whether the information presented is informative in regards to the firm’s financial performance, and about the relevance of the financial

performance and the ability of the accounting system to measure the financial performance. These definitions of earnings quality are all qualitative characteristics as described in paragraph 2.1, but are made more specific for earnings. This implies that if the earning information is relevant, represents faithful information, and is not lacking information in the other four quantitative characteristics, information in earnings is considered to be of high quality.

Even though decision usefulness is part of the definition of earnings quality in general (Dechow et al., 2010), investors are primarily looking for earnings persistence in firm

information. Earnings are considered to be persistent when income continues from one year to another. When earnings are more persistent, the decision usefulness of the earnings are higher (Dechow et al., 2010) and more predictable. Sloan (1996) found that earnings information in accruals are less persistent than the earnings information in the cash flow component, this suggests that firms with high level of accruals have a low quality of earnings (Dechow & Dichev, 2002).

The research of Ball & Shivakumar (2005) focused on earnings quality in the UK. They hypothesized that even though law and regulation are the same for both private and public firms, accounting quality is lower for private firms due to the difference in demand by both markets. By researching the timely loss recognition using a time series measure and an accrual based method Ball & Shivakumar (2005) conclude that the hypothesis they stated was correct. Market demands affect earnings quality. By researching the timely gain and loss recognition through accruals, Ball

(14)

& Shivakumar (2005) are able to give conclusions about earnings quality because a timely gain and loss recognition is based on not realized but expected cash flows (Ball & Shivakumar, 2005). These cash flows are presented in accruals.

Schipper & Vincent (2003) measure earnings quality by researching the persistence, the predictive ability, and the variance in time-series of earning by focusing on relations between cash, accruals and income. They use qualitative research to provide them with answers. Schipper & Vincent (2003) state that by researching earnings quality in this particular way, they will be able to give conclusions about relevance, reliability and comparability. Schipper & Vincent (2003) found evidence that the FASB considers the impact of standards on the variability of earning.

Measuring earnings quality can be done through various possibilities. Dechow et al. (2010) describe earnings persistence, abnormal accruals, earnings smoothness, timely loss recognition and target beating as parts of the overarching term of earnings quality. Earnings are persistent when investors can make accurate predictions about future valuation. This can be described as the usefulness of earnings for valuation. Accruals as a component of earnings are the most studied subject of persistence (Dechow et al., 2010). Abnormal accruals are components of accruals that are not in line with the expected cash flows (Wang, 2014). When investors can not predict expected cash flows well, it means that the accruals in the financial statements provide the investors with less useful information about future earnings, thus lowering the earnings quality. Different accrual models that are used to detect lower earnings quality will be described in paragraph 2.4.

Earnings smoothness is a proxy for predictability of future earnings (Welc, 2014). Investors are more interested in companies that have high earnings smoothness because high earnings smoothness means lower earnings instability (Welc, 2014). The smoothness of earnings is the outcome of an accrual-based system (Dechow et al., 2010) and it means that the company used accounting in such a way that fluctuations in the net income of the company will be leveled out. This leads to earnings that are predictable, thus catching the interest of more investors.

A timely loss recognition means that gains and losses are accounted for timely (Ball & Shivakumar, 2006). Timely loss recognition is also measured through accrual models. When loss recognition is not done timely, losses will be accounted for at a later time. This means that investors are making decisions based on numbers in the financial statements that do not contain the latest performance information of the firm, making the firms’ financial statements less useful.

Target beating is the term for beating targets set by analysts. An example of this is adjusting earning measures to report a small profit instead of a small loss (Dechow et al., 2010). The difference between a loss and a profit can be minimum, but the effects can be major.

(15)

Therefore research focused on earnings measure use small profits and small loss avoidance as indications of earnings management. The higher the earnings management the lower the earnings quality.

Thus, in order to draw conclusions about accounting- and research quality, earnings quality needs to be researched. And in order to draw conclusions about earnings quality, accruals need to be researched. As described in this paragraph, two important financial statement items to calculate the earnings persistence of a firm are cash flow and accruals (Bradshaw et al., 2001; Sloan, 1996). Accruals is the most studied component of earnings (Dechow et al., 2010) and is also the least earnings persistent item of the two components (Sloan, 1996). Due to the importance of accruals in this research, this subject is covered in the next paragraph.

2.4 Accruals

Accruals occur due to the existence of the matching principle in accounting rules. This matching principle is created in order to have costs and revenue allocated to the period to which they relate. For instance, insurance costs are often paid for in terms of one year. When the costs for 2015 are paid for in 2014, this costs have to be presented as accruals in the financial statements. Costs are not recognized in the year 2014. An accrual named ‘Pre-paid expenses’ is created and the contra account is the bank account. In 2015, the year the costs relate to, the costs will be recognized and the accrual will disappear. This is the matching principle in a nutshell. In order to further explain what accruals are and how they affect earnings, an example from Dechow et al. (2011) is presented in appendix 1.

Dechow et al. (2010) show that the definition of accruals has changed over time. Where in early research (Sloan, 1996; Jones, 1991) the accruals where defined as non-cash working capital and depreciation, research after the introduction of the Statement of Cash Flows define accruals more often as the difference between cash flow and earnings (Dechow et al., 2010). The main reason that accruals lacking earnings quality is because the accrual component of earnings is easier to manipulate in comparison to the cash flow component (Galanou, 2012). Manipulating the cash flow means cash will flow out of the organization, while using accruals does not lead to a decrease of cash. Therefore accrual component of earnings is less reliable than the cash flow component (Sloan, 1996; Leippold & Lohre, 2012). Managers can manipulate earnings to achieve a higher bonus by pulling earnings from next year to this year. This can be achieved by managers giving a higher discount percentage to clients. Clients will buy more and managers increase their bonus because there are more quantities sold.. This however leads to higher costs for the company because the discounts the managers give, should have been pure revenue. This is why

(16)

the accrual component is mostly used by managers. Galanou (2012) gives the example of understating current liabilities or prematurely record sales by managers to increase the bonus of that manager. Galanou (2012) also states that due to the complexity of this earnings management process, investors fail to recognize these manipulations.

In the research of Busman et al. (2015) researchers show that accruals can be used to provide higher earnings quality. This can be done by using accrual accounting systems.

Smoothening temporary timing fluctuations in operating cash flows is one of the fundamental attributes of accrual accounting, and by using accrual accounting systems, firms should produce higher quality earnings numbers by adding accruals to operating cash flows (Bushman et al., 2015). While appendix 1 shows how accruals occur and how they work, Bushman et al., 2015 shows that by using accrual accounting, economic events in firms can be recognized by reading the financial statements and by particularly focusing on the accruals in the financial statements of firms. As we will see further described in paragraph 2.5, this is not always the case.

The central prediction of the timing role of accrual accounting is that cash flows and accruals from operations are negatively correlated (Bushman et al., 2015; Dechow, 1994; Sloan, 1996). But even though research shows us this negative correlation is present, the research of Busman et al. (2015) also shows us that the overall correlation between cash flows and accruals has gone down from an adjusted R2_{of 70% in the 1960s up to nearly zero in recent years. Some} researches even state that the negative correlation has declined totally (Green et al., 2011). These negative correlation is known as ‘The Accrual Anomaly’ and will be described in paragraph 2.5.

To further understand how the earnings quality of accruals is lacking it is important to know the underlying components of these accruals, therefore we need to know how earlier research calculated their accruals. The research of Dechow et al. (2010) has been a great help here, as they research summarized the most used accrual models in their scientific research. In their research they discuss whether the models can represent distortions.

A still widely used model in scientific research is the Jones (1991) model. In this model, Jones (1991) defines the accrual process as a function of growth in sales and PPE. Even though Jones (1991) used the intuitive drivers of firm value in this model, the explanatory power of the Jones (1991) model is low and explains only around 10% of the fluctuations in accruals (Dechow et al., 2010).

The Jones (1991) model, as described in Dechow et al. (1995) is:

NDAt = α1 (1/At-1) + α2 (∆REVt) + α3 (PPEt)

(17)

NDAt = Nondiscretionary accruals;

∆REVt = Revenues in year t less revenues in year t-1 scaled by total assets at t-1; PPEt = Gross property plant and equipment in year t scaled by total assets at t-1; At-1 = Total assets at t-1; and

α1,α2, α3 = firm-specific parameters.

The estimations of the parameters are computed using the following formula:

TAt = a1 (1/At-1) + a2 (∆REVt) + a3 (PPEt) + υt.

Where:

TAt = Total accruals scaled by lagged total assets; and a1, a2, a3= Ordinary least squares of estimates of α1,α2, α3.

Due to a flaw in the Jones (1991) model, Dechow et al. (1995) adjusted the Jones model. The flaw in the Jones (1991) model is, that it assumes that revenues are nondiscretionary. A simple administration tool the management can apply is increase revenues and increasing

receivables. Dechow et al. (1995) used a modification within the Jones (1991) model to eliminate this flaw. The modified Jones model is:

NDAt = α1 (1/At-1) + α2 (∆REVt - ∆REC1) + α3 (PPEt)

Where:

∆RECt = Net receivables in year t less net receivables in year t-1 scaled by total assets at t-1. Even though Dechow et al. (1995) were modifying the Jones (1991) model to create a better model, Dechow et al (2010) state that the modified Jones model still suffers from Type I errors. A type I error means that accruals are classified as abnormal when they are in fact the right representation of the firms performance. Type II errors classify abnormal accruals as normal accruals.

Dechow & Dichev (2002) interpreted accruals different as in the two Jones models, because accruals require estimates and assumptions of future cash flows. This means Dechow & Dichev see accruals as a function of current, past and future cash flows (Dechow et al., 2010). This is in line with the example stated in appendix 1. The model Dechow & Dichev created and used is:

∆WC1 = b0 + b1CFOt-1 + b2CFO1+ b3CFOt+1 + εt

Where:

(18)

CFO = Cash flow from operations.

By applying this model, Dechow & Dichev (2002) focus on short term working capital accruals, while diminishing the long term accruals that affect the cash flows as well. This is a big limitation, knowing that impairments of PPE and goodwill are well known indicators for earnings management (Dechow et al., 2010). Therefore Francis et al. (2005) have created the FLOS model which modified the earlier model of Dechow & Dichev (2002). Frances et al. (2005) add PPE and growth to the model, items the previous model was lacking. However, they did not investigate whether these adjustments would decrease or even increase Type I and Type II errors (Dechow et al., 2010). Also, Francis et al. (2005) decomposed the standard deviation of the residual into firm-level measures (Dechow et al., 2010). In the model of Dechow & Dichev (2002) it was not possible to state conclusions about adjustment of accruals due to managerial choices, in the new model the change into firm-level measures provides research with this option. The model of Francis et al. (2005) which was already proposed by McNichols (2002) is:

TCAt = α + β1CFOt-1 + β2CFOt + β3CFOt+1 + β4∆REV + β5PPEt + εt

Where:

TCA = Total current accruals;

∆REV = Firms changes in revenues between year t-1 and year t; and PPE = Firms depreciation and amortization expense.

Sloan (1996) was not focused on earnings management, but on the effects of accruals on the investors’ decisions. Sloan (1996) computed accruals as:

Accruals = ( ΔCA – ΔCash ) – ( ΔCL – ΔSTD – ΔTP ) – Dep

Where:

ΔCA = Change in current assets;

ΔCash = Change in cash/cash equivalents; ΔCL = Change in current liabilities;

ΔSTD = Change in debt included n current liabilities; ΔTP = Change in income taxes payable; and Dep = Depreciations and amortization expense.

This formula is consistent with the definition of accruals during the research of Sloan (1996). Galanou (2012) in her research, shows, next to the accrual formula stated by Sloan (1996) also another formula:

(19)

Accruals = ( ) − ( ) − Depreciation

This formula is more in line with the second definition of accruals, being non-cash working capital less depreciation (Galanou, 2012). Now that is explained what accruals are, how they occur, how accruals and earnings management can be measured and why accruals provide difficulties for investors to estimate future earnings, we can zoom in on the phenomenon that is caused by investors that do not react to this accruals, the accrual anomaly.

2.5 The accrual anomaly

The accrual anomaly is the name for an effect in the economy which occurs when investors do not understand accruals of firms. Due to this misunderstanding, stock prices are overvalued or undervalued, which contradicts the efficient market hypotheses that states that all publicly available information effect the stock prices, thus creating an efficient market. Therefore these stock prices seem to lag information, thus creating the accrual anomaly.

The accrual anomaly, which is in fact, not an anomaly at all (Dechow et al., 2011) was detected for the first time by Sloan (1996). The accrual anomaly is not an anomaly because an anomaly is a behavior that deviates from an existing theory, and at the time Sloan (1996) found the accrual anomaly, Sloan (1996) was testing a well-known and well-supported theory (Dechow et al., 2011). At the time of the research of Sloan (1996), Ou & Penman (1989) and Bernard & Thomas (1990) had paved the path in terms of predicting future earnings by respectively showing how to predict future earnings one year ahead (Ou & Penman, 1989) and by showing how to predict future quarterly earnings (Bernard & Thomas, 1990). Wilson (1987), Bernard & Stober (1989) and Lev & Thiagarajan (1993) had caught the interest of Sloan (1996) by

investigating the cash flow and accrual components of earnings by focusing on stock returns. This research would not be that important if Sloan (1996) did not combine the knowledge of the researches of Bernard & Stober (1989), Bernard & Thomas (1990), Lev & Thiagarajan (1993), Ou & Penman (1989) and Wilson (1987), in order to expose the accrual anomaly. But the first hint of the existence of the accrual anomaly was already given a long time ago by Graham & Dodd (1934) (as cited in Dechow et al., 2011, p. 1). Due to the at that moment new theory known as the efficient market hypothesis, and due to academics stating that the coherence between stock prices and accounting earnings were good because earnings were good at showing the intrinsic value of a firm (Dechow et al., 2011) the academics didn’t believe in the existence of the accrual anomaly .

(20)

Sloan (1996) hypothesized that the accrual component of earnings was of lower quality than the cash flow component of earnings (Dechow et al, 2011). By computing earnings, accruals and cash flows for data from the US between 1962 and 1991 and creating portfolios by sorting the companies within the dataset from highest accruals to lowest accruals for one fiscal year, Sloan (1996) analyzed the earnings information of that fiscal year. Sloan (1996) combined the earnings information of one fiscal year with the earnings information of the returns that are gained in the year t+1. Results of his data analysis show that by reacting on accruals when high or low, abnormal returns up to 10,4% can be achieved. This return can be explained as over-/undervaluation of firms (Sloan, 1996). Sloan (1996) showed this return of 10,4% can be made by taking a long position in the stock of firms that report low levels of accruals and by taking a short position in stock of firms that report high levels of accruals. This 10,4% is created by a profit of 4.9% in long positions and a profit of 5.5% in short positions. By creating a hedge, 10,4% can be made. In the conclusion of his research Sloan (1996) gives a possible explanation for his findings, stating that this might be “evidence of a normal return to an active investment strategy based on financial statement analysis.” (p. 314). But overall his research was an appeal to future research, which was widely heard by researchers all over the world.

These researchers all focused on the existence of accrual anomaly, but linked the accrual anomaly to different statistics, like investors (Battalio et al., 2012; Bradshaw et al., 2001; Collins et al., 2003; Galanou, 2012; Hirshleifer et al., 2012; Hiba & Yehuda, 2015; Momente et al., 2015; Pincus et al., 2007), analysts (Barth & Hutton, 2004; Bradshaw et al., 2001; Mohanram, 2014), and equity (Kothari et al., 2006; Sawicki & Shrestha., 2011). Due to the approach of this research, my literature study is mostly focused on investors. By recapitulating the influential research, the hypotheses will be formed in the next chapter..

Bradshaw et al. (2001) focused their research on the information that analysts and accountants provide to the investors in relation to future earnings of companies. Bradshaw et al. (2001) found evidence that analysts do not inform their readers with the information necessary to respond to accrual information and they found no evidence that accountants provide investors with information about the future earnings problems. The results regarding the reporting of the accountants suggest that financial statements are lacking reporting quality. The conclusion of this research supported the results of Sloan (1996). Bradshaw et al. (2001) showed that investors do not fully react to accruals and accrual information about future earnings, thus rejecting the efficient market hypothesis.

Battalio et al. (2012) researched who reacts to accrual information by using different groups of investors. Battalio et al. (2012) found that the majority of investors ignore accrual

(21)

information at the moment this information becomes public. By researching different groups of investors based on the amount of shares bought and sold, Battalio et al. (2012) found that investors that initiate trades of at least 5.000 shares seem to react in the proper direction, thus interpret the accrual information correctly and utilizing the accrual information right. But this research also provides evidence that investors that initiate in the smallest trades appear to respond to “attention grabbing” stocks, and react in the opposite direction of the accrual information. Due to these two findings this research showed that investors that take the accrual information in to account while trading, have insufficient market power to mitigate the accrual anomaly. This implies that although the smallest investors seem to be the cause of the accrual anomaly, this group is big enough to maintain this phenomenon.

Collins et al. (2003) primarily looked at the role of institutional investors in relation to pricing of the accruals. Collins et al. (2003) finds evidence that firms that have a high level of institutional ownership and a minimum level of active institutional traders show more accurate stock prices and reflect the persistence of accruals more accurately. These findings tell us that firms with a high level of institutional investors should not have huge overpriced or underprices stock. With these findings Collins et al. (2003) suggest that institutional investors can more accurately price the stocks, therefore finding that a certain group of investors do not seem to ignore the accrual information needed to quickly react to future earnings information and suggesting that the smaller investors are mispricing shares and stock.

Galanou (2012) focused on the Dutch stock market. Galanou (2012) found evidence that the cash flow component in comparison with the accrual component of current earnings is significantly more persistent. However, while using a similar method Sloan (1996) used to show that abnormal returns can be earned, Galanou (2012) did not find evidence that abnormal hedge returns can be earned by investing in a long position in low accrual firms as well as investing in a short position in high accrual firms. But the findings of Galanou (2012) did suggest that

investors do not understand the persistence of accruals fully which leads to undervaluation or overvaluation of shares.

Hirshleifer et al. (2012) researched the underlying cause of the accrual anomaly,

delimiting this research by suggesting that this can be caused either by risk that investors take or mispricing of the accruals thus mispricing stock by investors. The findings suggest that investors misvalue accrual characteristics, which is consistent with Sloan (1996) and Bradshaw et al. (2001).

Research of Hribar & Yehuda (2015) is fixated on the accrual anomaly in different stages of firms. After testing whether the accrual anomaly exists, the tests were focused on different stages of firms. Hribar & Yehuda (2015) find that the correlation between the total accruals and

(22)

the free cash flow grows as a firm matures. They state that both the free cash flows as well as the accruals are mispriced “to the highest degree” (p.225) when a firm is in its growth stage. This means that investors are failing to predict the future earnings of a company in their growth phase.

Momente et al. (2015) researched the company’s future performance by focusing on accruals and on whether the information about these future performances in accruals can be attributed to risk. By researching accruals and future stock prices, findings of Momente et al. (2015) suggest that the lower future stock returns are more likely to be the result of investments decisions by investors rather than risk avoidance by the firms. These findings suggests that investors and analysts do not fully understand the accruals and do not fully know the potential future earnings of a firm, resulting in undervaluation of the firms.

While all earlier research fully focused on the different type of investors, Pincus et al. (2007) tried to find international evidence for the accrual anomaly. Therefore they did not primarily focus on investors but this research focused on the existence of accrual anomaly within different countries. Where most of the research use data from the U.S., Pincus et al. (2007) dataset consists twenty different countries, containing data from the Dutch market. Pincus et al. (2007) find that the accrual anomaly is most likely to occur in common law countries, in

countries that allow extensive use of accrual accounting and in countries that have a less share ownership. But the finding Pincus et al. (2007) were mostly surprised about was finding the accrual anomaly in countries where the market is considered to be most efficient. However overall they find that the relationship between the fundamental investor information and the stock price of companies is weaker in less efficient markets.

While Sloan (1996), Bradshaw et al (2001) and Hirshleifer et al (2012) suggest that not all investors react in a correct way to accrual information, Collins et al. (2003) and Battalio et al. (2012) show that the institutional investors tend to respond correctly to the public accrual information, thus reacting accurately to future earnings information. Even though all of the above mentioned research is focused on different perspectives and they are all running different tests, each of these papers shows us that accrual anomaly, +/- 20 years after the research of Sloan (1996), is still around. These findings suggests that investors are still not aware of the future earnings problems that accruals may cause.

However, this is not the explanation Kraft et al. (2006) showed us. Kraft et al. (2006) investigated the different theories and explanations for the accrual anomaly. By performing robustness test to exclude small numbers of firm-year observations, Kraft et al. (2006) find evidence that accrual anomaly is very unlikely to be a result of the inability of investors to

(23)

understand accrual information. Kraft et al. (2006) actually state that by performing no robustness check by excluding small numbers of firm-year observations, it is very easy “for researchers to generate “false positives” for hypotheses related to market mispricing” (p. 332). Kraft et al. (2006) specifically used the robustness check on the research of Sloan (1996) and they conclude that in this research the accrual anomaly is unlikely to be caused by the investors’ inability to understand accrual information of firms.

Also, Green et al. (2011) show that the accrual anomaly is not as strong as it was during the research of Sloan (1996). Green et al. (2011) find evidence that the accrual anomaly had decreased in the U.S. stock market. The researchers even use the words ‘decayed’ and ‘demise’ suggesting that accrual anomaly is gone, but they state that their findings do not support the findings of Kraft et al. (2006). Green et al. (2011) reported these results by finding evidence using the method of Sloan (1996) in which he shows that hedge returns up to 10,4% can be made. Green et al. (2011) used this method and showed that as of today, anticipating on the stock market with the Sloan-method this 10,4% return is no more. They also find evidence that returns with this method can be negative.

Even though Bushman et al. (2015) did not use the terms ‘decayed’ and ‘demise’, they also state that the negative association between accruals an operating cash flows has disappeared nearly by finding strong evidence that the correlation between cash flows and accruals has declined significantly. Despite the fact that the findings of Bushman et al. (2015) show that the correlation declined, they cannot give a clear answer to why this correlation diminishes.

But the findings of Green et al. (2011) however are not consistent with the findings of Lev & Nissim (2006). Where Green et al. (2011) speaks of the disappearance of accrual anomaly, Lev & Nissim (2006) find evidence about accrual anomaly being persistent. In their research they show that accrual anomaly is still present, years after the research of Sloan (1996), but also that the magnitude has not declined over time which contradicts the findings of Bushman et al. (2015).

These papers show us that a lot of research has been done to understand accrual

anomaly and the persistence of this phenomenon, and that most of the papers state that accrual anomaly is still very alive after all those years.

(24)

3 Methodology

3.1 Methodology

Where Battalio et al. (2011) focused on the differences in stock prices and earning

announcement, I want to focus on the financial statements and the future earnings that financial statements provide the investors which. For this research the methodology Sloan (1996) used is a good basis. Where Sloan (1996) made hedge portfolios to show that anticipating on high or low accruals is very profitable, I want to link this research to the volume of stock traded in order to find evidence that answer the research question. Therefore my first hypothesis is identical to H2(ii) of Sloan (1996):

H1 : A trading strategy taking a long position in the stock of firms reporting relatively low levels of accruals and a short position in the stock of firms reporting

relatively high levels of accruals generates positive abnormal stock returns. Sloan (1996) used the H1 hypothesis as an extension of his research to provide evidence that reaction to accrual information can provide the investors with high profits. I will use this hypothesis in order to test whether years after the first test Sloan (1996) showed the world, I will find the same evidence Sloan (1996) provided. The difference is that this research will be using data up to 2014 and that this research will zoom into the Dutch Stock market in order to show how investors in this market react to earnings-information in financial statements. By using data from the stock market this research will connect the amount of shares traded with the abnormal returns as provided by testing H1. As stated in previous literature research, the institutional investors seem to react in the right way to future earnings information in the financial

statements. Institutional investors are million/billion dollar companies and willing to invest more money than small investors, thus trading more volumes of shares. Therefore in this research the hypothesis is formed that companies in which the most shares are traded, show that the accrual anomaly is lower or inexistent. This will be tested by using the following hypothesis:

H2 : Accrual anomaly is more likely to be found in companies where a small amount of shares are traded in contrast to the companies where a lot of shares are traded. After H2 is tested and the results between the data with the small trading volumes and the high trading volumes is available, this research focusses on the lowest and the highest accrual

(25)

portfolios in both groups in order to find evidence not only on which group of investors cause the accrual anomaly to exist, but also to provide evidence on which investment choices are the leading cause of the accrual anomaly in the Netherlands. For these tests the following hypothesis are formed:

H3a : Abnormal returns in companies with low accruals, where small amounts of shares

are traded are lower than the abnormal returns in companies with low accruals where high amounts of shares are traded.

H3b: Abnormal returns in companies with high accruals, where small amounts of

shares are traded are lower than the abnormal returns in companies with high accruals where high amounts of shares are traded.

Where sophisticated investors are mostly well funded and trading big amounts of stock, the smaller investors are mostly trading in less amount of stocks and mostly do not have the information that sophisticated investors have (Battalio et al., 2012; Collins et al., 2003). Therefore the data about the small amounts of shares trades will be treated as being data of smaller

investors, and the data about the high amounts of shared trades will be treated as being data of institutional investors.

3.2 Sample & Variable Measurement

The sample consists of data from 2005 up to the end of the fiscal year of 2014, extracted from DataStream. This data contains financial statement information about AEX listed companies as well as information about the amount of shares traded in a particular month in the time period stated before. The dataset that contains the financial statement information consists of 1.056 firm years. These fiscal years are represented by 116 unique companies. This is relatively small dataset compared to datasets from the US which is caused by the smaller size of the Dutch Stock market. Financial variables of interest, consistent to the research of Sloan (1996), are earnings, accruals and cash from operations. To test the formed hypotheses, first of all the accruals of the companies within the data have to be computed. Here accruals are computed by using the formula also used in Sloan (1996), due to the overlap in tests between the research of Sloan (1996) and my research. Accruals are computed as follows:

(26)

The variables in this formula are: ∆CA = Change in current assets;

∆Cash = Change in cash/cash equivalents; ∆CL = Change in current liabilities;

∆STD = Change in debt included in current liabilities; ∆TP = Change in income taxes payable; and

Dep = depreciation and amortization expense.

Income taxes payable is excluded from accruals to have consistency with the earnings that are deployed in the empirical tests. Debt in current liabilities is also excluded from accruals because this does not relate to operating transactions but to financing transactions (Sloan, 1996). The Cash Flow component is computed as the difference between earnings and the accrual component of earnings.

This research starts off by computing the different components of earnings, as described in Sloan (1996), Dechow et al. (2011) and Galanou (2012). The different components are

computed as follows:

Earnings = ;

Accrual Component = ; and

Cash Flow Component = .

The dataset that contains the stock information consists of 14.535 fiscal months and is obtained from DataStream. This dataset contains information about the volume of stock traded, as well as abnormal return data.

Like the research of Sloan (1996) stock returns are computed for three future years. This means that for every year there are earnings, divided into a cash flow component and an accrual component, and behind this information we see what the abnormal returns for this company are for the year+1, for the year+2 and for the year+3. This data will be divided into 5 different portfolios, from low accruals (group 1) to high accruals (group 5). Sloan (1996) uses 10

portfolios. 10 portfolios however, is not optimal in this research due to the smaller sample size. The data ultimately is pooled in order to test the first hypothesis.

(27)

3.3 Methodology

This part of the research will describe the steps taken in the data analyses.

For H1, only the dataset with the financial statement information is used, starting with the years 2005 up till 2014, containing 1.056 fiscal firm years. The data that has been removed from the sample contains data of the full year of 2005, data of the full year of 2014 and the companies with missing data within the years 2006 up till 2013. Due to 2014 being a year where not all the data was yet available at the moment the data was retrieved, 2010 is the last year abnormal results for three upcoming years could be computed. 2005 is removed from the sample because it lacked the data of the first month of 2005. For each fiscal year for each company the earnings, accrual component of the earnings, cash flow component of the earnings and the abnormal returns of the year+1, abnormal returns of the year+2 and the abnormal returns of the year+3 are computed. Then, the total of the companies are divided by five in order to create five portfolios from low accruals up to high accruals.

In order to answer the first hypothesis, the results for all the years are pooled and the two-tailed t-test is performed on the abnormal result data of the portfolios with the lowest accruals during the years 2006 till 2010, and on the abnormal result data of the portfolios with the highest accruals during the years 2006 till 2010. The pooled sample contains 114 data points.

For H2, the stock market dataset is tested. Due to the research in the first dataset, in this dataset only the years from 2006 up till 2010 can be used. The dataset contains of 12 data points (months) during the years 2006 till 2010 and shows the amount of stock/shares traded in that day. First the data of a firm that consists of 12 months have to be computed into yearly data. Therefore the volumes are added and divided by 12 in order to compute the average stock volume that is traded for a company in one year. Then for every year a check is performed to see if the companies are also present in the sample of H1. Companies without a match with the dataset used in H1 are excluded. Hereafter the total averages of the volume data were added and the average was computed. After this calculation every company is dealt into a specific group. The first group is the group with the lowest amount of shares traded, here the average stock volume of the company is lower than the average stock volume of the total dataset for the year. The second group is the group with the highest average stock volume. This group contains considerably less companies than the first group because the Dutch stock market contains a few big companies and a lot of small companies. The companies that show the highest stock volume, are companies such as Heineken, Unilever, KPN, Shell, Ahold, Philips and Air France-KLM.

(28)

After the volume data for the years are computed and the companies are divided into the two groups, the same test as in H1 is conducted. To conduct the test the dataset from H1 is combined with the dataset of H2.

After conducting the test, the first group, the group with the lowest volume of shares traded is divided into 5 portfolios from lowest accruals to highest accruals and a two-tailed t-test is performed on the future earnings of the companies with the highest accruals and the lowest accruals in this portfolio.. In the second group, the group with the highest volume of shares traded the total group is divided into 3 portfolios due to the small amount of data in this group. In the second group the two-tailed t-test is performed on the lowest accrual portfolio and the highest accrual portfolio. By performing two-tailed t-tests within the created groups of small investors and institutional investors, it is possible to answer the hypothesis stated in H2.

For the test of H3 the same data as used in H2 is used. To test this hypothesis, the group that has the lowest accruals and the lowest amount of shares traded is compared with the group with the lowest accruals and the highest amount of shares traded. Also the group with the highest accruals in the group of the lowest amount of shares traded is compared to the group with the highest accruals in the group of the highest amount of shares traded. This test is performed in order to see if there are significant differences between the small investors and the institutional investors. Also by conducting this test, it is not only possible to point out which type of investors cause the accrual anomaly to still exist, but also to show which investment choices cause the accrual anomaly to exist, providing science with new insights.

(29)

4 Empirical Results

4.1 Testing H1

For the first hypothesis, the earnings data has to be combined with the abnormal return data as the hypothesis is:

H1 : A trading strategy taking a long position in the stock of firms reporting relatively low levels of accruals and a short position in the stock of firms reporting

relatively high levels of accruals generates positive abnormal stock returns. As mentioned in chapter two, investors seem to have problems to distinguish the accrual component and the cash flow component from earnings. Sloan (1996) showed that trading on the stock market can generate abnormal returns when understanding the different components of earnings. As Sloan (1996) stated :”In particular, a long position in firms reporting low levels of accruals relative to cash flows and a short position in firms reporting high levels of accruals relative to cash flows should yield positive abnormal stock returns.” (p.306). Sloan (1996) also showed in his research that creating a hedge portfolio of a long position in the lowest accrual portfolio and a short position in the highest accrual portfolio, a return of 10.4% could be made. Table 1 shows the outcomes by testing the first hypothesis in the Dutch stock market for fiscal firm years between 2006 and 2010. The results of test H1 as shown in table 1 do not confirm that accrual anomaly is still at the same level as it was during the time of Sloan (1996). Column 2 in Table 1 show that in year t+2 a significance at the 0.10 level is found. These results are contradictive to most of the research stated in chapter two (Battalio et al., 2012; Bradshaw et al., 2001; Galanou, 2012; Pincus et al., 2007) but is more in line with the research of Green et al. (2011). Green et al. (2011) showed us that the accrual anomaly in the US is not at the same level as it was back in 1996 and they suggest even that the accrual anomaly is gone. Green et al. (2011) even show that returns based on the method of Sloan (1996) can result in negative earnings. This is also the result of this research, shown in Table 1. Where Sloan (1996) showed a hedge of 0,104 with a significance at the 0.01 level in the year t+1, this research shows a hedge of -0,007,

insignificant and therefore no accrual anomaly is found in year t+1. However, year t+2 shows a hedge of -0,034, significant at the 0.10 level. Results of year t+3 show a hedge of 0,001,

insignificant. The result of t+2 suggests that investing in the opposite direction to the research of Sloan (1996), a result of 3,4% can be gained. This means that in order to get these results, a short position in firms that show low levels of accruals and a long position in the firms that show high levels of accruals is the way of gaining profits. The accrual anomaly is still present, but in the

(30)

small Dutch stock market in comparison to the US stock market, results of H1 show that the accrual anomaly works in the opposite direction.

Portfolio Acrual

Ranking year t+1 year t+2 year t+3

Lowest -0,021 -0,009 -0,011 2 0,003 -0,020 -0,012 3 0,008 0,019 -0,011 4 0,004 0,017 0,007 Highest -0,014 0,025 -0,013 Hedge (c) -0,007 -0,034 0,001 P - Value ( 0,737 ) ( 0,084 ) ( 0,941 )

b. The abnormal returns are obtained by AEX data in year t+1.

Table 1

c. The hedge portfolio consists of a long position in the lowest accrual portfolio and an offsetting short position in the highest accrual portfolio

Portfolios are formed annually by assigning firms into deciles based on the magnitude of accruals in year t.

Table 1

Time-series Means of Equal Weighted Portfolio Abnormal Stock Returns Sample consists of 287 Firm-years between 2006 and 2010 (a)

Abnormal Returns (b)

a. Accruals: The change in non-cash current assets, less the change in current liabilities (exclusive of shortterm debt and taxes payable), less depreciation expense, all divided by avarrage total assets.

The first hypotheses can be rejected due to the significant result that shows that a trading strategy taking a long position in the stock in the stock of firms reporting relatively low levels of accruals and a short position in the stock of firms reporting relatively high levels of accruals generates negative abnormal stock returns instead of positive abnormal stock returns.

4.2 Testing H2

The first hypothesis has resulted in a rejection of the hypothesis. In H2 the data of H1 is separated in small investors and big investors. Battalio et al. (2012) showed that institutional, big investors tend to react in the right way when it comes to understanding accruals. Smaller

(31)

investors tend to look at the numbers at the bottom of the profit and loss statements. Because this group of small investors is so big, the accrual anomaly still exists. H2 is stated as follows:

H2 : Accrual anomaly is more likely to be found in companies where a small amount of shares are traded in contrast to the companies where a lot of shares are traded. To answer this hypothesis research started focusing on the smaller investors. Results in table 2 show no significant results. Therefore the results of this research do not support the results of Battalio et al. (2012). This sample consists of 214 firm-years between 2006 and 2010. These are all firms in which the volume of stock trades is less than the average stock traded. This means that the majority of the Dutch stock market consists of smaller companies.

Portfolio Acrual

Lowest -0,021 0,001 -0,013 2 -0,005 -0,014 -0,007 3 0,020 0,029 -0,013 4 0,011 0,015 0,019 Highest -0,010 0,010 -0,018 Hedge (c) -0,010 -0,009 0,005 P - Value ( 0,670 ) ( 0,688 ) ( 0,784 )

Table 2

Table 2

Time-series Means of Equal Weighted Portfolio Abnormal Stock Returns Sample consists of 214 Firm-years between 2006 and 2010 (a)

(32)

Table 3 shows the institutional investors, the companies in which the highest volumes of stock are traded. This sample consists of 55 firm-years between 2006 and 2010. This is a small sample because there are just a few companies in which trading volumes are extremely high compared to the smaller companies. As also H1 shows, the significant results are presented in year t+2. Table 3 shows a significance at the 0.01 level and the hedge shows that creating a hedge between a short position in low level of accrual firms and a long position in high level of accrual firms can result in positive stock gains of 6,9%. Even though this test shows a significant result at the 0.01 level, due to the sample size this result can be questioned.

Portfolio Acrual

Lowest -0,012 -0,045 -0,022

2 -0,006 -0,002 -0,004

Highest -0,000 0,024 -0,013

Hedge (c) -0,012 -0,069 -0,009

P - Value ( 0,644 ) ( 0,002 ) *** ( 0,664 )

*** Denotes significance at the 0.01 level using a two-tailed t-test Table 3

Table 3 Large investors

Time-series Means of Equal Weighted Portfolio Abnormal Stock Returns Sample consists of 55 Firm-years between 2006 and 2010

H2 is stated in a way that suggests small investors to cause the accrual anomaly, consistent

with Battalio et al. (2012) and Collins et al. (2003). The results however show that the accrual anomaly which is found in H1 is not present due to the investment choices made by the small investors, but is present due to institutional investors. Therefore H2, like H1 is also rejected, and the findings are inconsistent with findings in earlier studies (Battalio et al., 2012; Collins et al., 2003; Hadani et al., 2011; Malmendier & Shanthikumar, 2007; Shanthikumar, 2012).

Understanding Accruals, Researching investors in Dutch listed companies : the blindness of investors